The Predictors of Surgery for Symptomatic, Atraumatic Full-Thickness Rotator Cuff Tears Change Over Time: Ten-Year Outcomes of the MOON Shoulder Prospective Cohort

Abstract

Background: A prospective cohort study was conducted to assess the predictors of failure of nonoperative treatment, defined as the patient undergoing surgery for symptomatic, atraumatic full-thickness rotator cuff tears. We present the 10-year follow-up data of this population to determine if predictors for surgery change over time, and secondarily we report the outcomes of the cohort.

Methods: At the time of enrollment, demographic, symptom, rotator cuff anatomy, and patient-reported outcome data were collected in patients with symptomatic, atraumatic full-thickness rotator cuff tears. Patients underwent a standard physical therapy protocol for 6 to 12 weeks. Patient data were then collected at 1, 2, 5, 7, and 10 years. Failure of nonoperative treatment was defined as the patient electing to undergo surgery.

Results: Of the 452 patients in the original cohort, 20 patients (5%) withdrew from the study, 37 (9%) died before 10 years, and 40 (9%) were otherwise lost to follow-up. A total of 115 patients (27.0%) underwent a surgical procedure at some point during the 10-year follow-up period. Of these patients, 56.5% underwent surgery within 6 months of enrollment and 43.5%, between 6 months and 10 years. Low patient expectations regarding the efficacy of physical therapy were found to be a predictor of early surgery. Workers' Compensation status and activity level were more important predictors of later surgery. Patient-reported outcome measures all improved following physical therapy. For patients who did not undergo a surgical procedure, patient-reported outcome measures did not decline over the 10-year follow-up period.

Conclusions: Low patient expectations regarding the efficacy of physical therapy were found to be a predictor of early surgery, whereas Workers' Compensation status and activity level were predictors of later surgery. Physical therapy was successful in >70% of patients with symptomatic, atraumatic full-thickness rotator cuff tears at 10 years. Outcome measures improved with physical therapy and did not decline over the 10-year follow-up period.

Level of evidence: Prognostic Level I . See Instructions for Authors for a complete description of levels of evidence.

Fig. 1

Ten-year Kaplan-Meier curve showing survival free from operative treatment among patients with symptomatic, atraumatic full-thickness RCTs. Overall, <30% of patients had undergone surgery by the end of the 10-year follow-up period. Most patients who underwent surgery did so within 6 months (red line).

Fig. 2

Predictors of early surgery. This graph reflects the characteristics of patients who underwent surgery within 6 months of enrollment. The strength of each predictor is reflected by its position on the x axis. Predictors to the right of the vertical red line were significant (p < 0.05). Significant predictors of early surgery include smoking, Workers’ Compensation status, the amount of tear retraction (see Figure 4), activity level, and patient expectations. The strongest predictor in this group, by far, was the patient’s expectations regarding the effectiveness of physical therapy.

Fig. 3

Predictors of later surgery (6 months to 10 years). This graph reflects the characteristics of patients who underwent surgery after 6 months of enrollment. The strength of each predictor is reflected by its position on the x axis. Predictors to the right of the vertical red line were significant (p < 0.05). Significant predictors of later surgery include patient expectations, activity level, and Workers’ Compensation status. Workers’ Compensation status and activity level were stronger predictors of later surgery than patient expectations regarding the effectiveness of physical therapy. df = degrees of freedom.

Fig. 4

The influence of RCT retraction on predicting early surgery. RCT retraction was a significant predictor of early surgery. The effect of retraction is adjusted for all covariates in the model. Interestingly, larger tears were less likely to be operatively treated than smaller tears.

Fig. 5-A

Fig. 5-B